https://github.com/osquery/osquery logo
Join Slack
Powered by
I clicked some things around what actions are requ...
# core
s
I clicked some things around what actions are required. And https://github.com/osquery/osquery/pull/6953 should remove Azure.
s
I'm not sure where these come from specifically but it seems that they'll never become "ready"
I would say that the required checks on all the specific workflow instances are enough
s
Those might have been azure ones. I don’t know if I clicked all the right things
They might also be older github actions
s
yeah I mean, they match the names of the workflows on github actions, but they don't seem to be used
s
I have unchecked them as required
This might mean we’re tracking something off in the workflows. But we can learn and iterate
From a UX perspective, it feels like we’re missing something. Like we should be able to require the workflow to pass, and not enumerate the sub-OSes in the workflow. So I think we’re probably missing something
s
I think this was noticed by Alessandro too initially when adding the support. There were no checks for the whole thing
while for instance Azure provides a check for the entire pipeline
s
Yeah, I remember that about your Azure setup. It seemed nice.
A quick google finds https://github.community/t/status-check-for-a-matrix-jobs/127354/2 which is a way to flag it
s
Right, it needs to be created manually. Sorry I actually remember talking about this too now. Good catch though.
s
I don’y think it’s a blocker. I think it’s mostly a nice to have to correctly manage the required checks on a branch
Anyhow, I removed the one that weren’t happening.
s
It is still partially annoying because we have multiple jobs and we need to list them in this final job we use for the check, while on Azure it was automatic.. you couldn't forget to add one.
s
Maybe. There might be a way to create a single one that farms it out. But I think we’re also just shuffling it around.
Either we have a single job that depends on the matrix jobs. Or we have a success job per matrix.
s
yeah but in every case you need to change the build.yml file to make it work. Just saying that's different, beyond that, we will live! ahah
s
I don’t think we do
the link above had some suggestions for how to depend on the entire matrix
s
There are multiple matrix
one for each platform
and also jobs that are outside the matrix logic
s
Right. We need a job per matrix, one a job that enumerates all matrixes.
While we might change what’s in the matrixes, I don’t think we add new platforms very often
s
Well there are also jobs which are "standalone", like the check source code one
s
Sure. But we’re still talking 4 — lint, plus the 3 platforms
s
Just to clarify I'm not saying that it's a major issue, just saying that it's not as automatic as it was on Azure which was automatically adding whatever job was created under the single "osquery" pipeline check. We definitely have ways to make it work in GA just saying that if you have to change the build.yml to add a new standalone job, you've got to remember to add that to the final check job too, unless there's a way to list all the jobs into the 
needs
key, programmatically.
s
Oh, I see… in Azure you could do a single “everything passes”
Yeah, perhaps not as clean
s
yeah sorry if I made it seem bigger than it was ^^'
s
As an aside, it’s kinda weird comparing Azure to github. Given that the latter is built on the former, I’d expect more convergence. But they’re oddly different
s
yep ahah. Probably different UX/UI teams
(just random guessing, it's not necessary obv.)
3 Views
Open in Slack
PreviousNext
Powered by